Software Engineer, ML Training Performance
Core
Design, implement, and operate cutting-edge ML training performance optimization techniques to scale VLM, VLA, and Foundational models training efficiency for autonomous driving.
Role type
Senior IC ML training performance engineer
Builds
Large-scale ML training infrastructure and RL frameworks for autonomous driving
Domain
Autonomous driving / Machine Learning Infrastructure
Deliverable
production ML models
Required skills
Distributed training frameworks, GPU/accelerator cluster optimization, RL frameworks, collective communication libraries, profiling tools, Python/C++, ML compilers, hardware/software co-design
Preferred skills
None stated
Technologies
Megatron-LM, TorchTitan, Verl, MILES, NCCL, NVIDIA Nsight, PyTorch Profiler, XLA, JAX
Responsibilities
Design and implement distributed training techniques (data/tensor/pipeline parallelism, mixed-precision, gradient checkpointing, GPU kernels); Work on RL framework and performance optimization; Profile and debug training throughput bottlenecks; Collaborate with researchers and hardware engineers; Implement custom kernels and build benchmarks
Seniority
Senior, hands-on IC